Evaluation of Sampling and Concentration Methods for Salmonella enterica Serovar Typhi Detection from Wastewater

Organism culture and enumeration.

Two commonly used strains of Salmonella Typhi were used as positive controls in this study, Ty21a and Ty2. Salmonella Typhi strain Ty21a is used in the oral, live attenuated typhoid vaccine and is negative for the Vi antigen. Salmonella Typhi strain Ty2 is a well-characterized, reference strain that is positive for the Vi antigen. To confirm the presence of the Vi antigen in Ty2 prior to experiments, a Vi antigen agglutination test was performed using Difco™ Salmonella Vi Antiserum (Becton, Dickinson and Company, Franklin Lakes, NJ). Salmonella enterica serovar Typhimurium (Salmonella Typhimurium) was used as a negative control in the agglutination test. Ty21a (33459; American Type Culture Collection, Manassas, VA), Ty2, and Salmonella Typhimurium were obtained from Dr. Stephen Libby (University of Washington). For each strain, 10 µL antiserum and 10 µL overnight culture were combined on a glass slide and examined for agglutination after 15 minutes. Agglutination indicated the presence of the Vi antigen in Ty2.

Throughout this work, Ty21a was grown using LB-Miller broth (IBI Scientific, Dubuque, IA), and Ty2 was grown in the dark using LB-Miller broth with a supplemental aromatic amino acid mix and 50 ng/mL ferrioxamine E (Millipore, Burlington, MA). The aromatic amino acid mix was prepared in a 100× stock consisting of L-Phenylalanine (4 mg/mL) (TCI America™, Portland, OR), L-Tryptophan (4 mg/mL) (Acros Organics, Fair Lawn, NJ), 2,3-dihydroxybenzoic acid (1 mg/mL) (TCI America), and para-aminobenzoic acid (1 mg/mL) (Sigma-Aldrich, St. Louis, MO), which were dissolved in deionized water and filter sterilized. To improve our understanding of Ty21a and Ty2 growth and therefore ensure that experiments were seeded during exponential growth, growth curves were determined for Ty21a and Ty2 and measured via optical density at 600 nm and spot plating of 100 µL of relevant dilutions³⁹ on LB-Miller agar or LB-Miller agar with a supplemental aromatic amino acid mix and 50 ng/mL ferrioxamine E. Ty21a and Ty2 inocula for experiments were prepared by growing the organisms for a specified period of time at 37°C, monitoring for exponential growth phase, and storing single-use aliquots of the cultures in 30% glycerol until use (−80°C). Prior to planned experiments, 20 µL of the frozen Ty21a or Ty2 glycerol stock was inoculated in 15 mL of liquid media and incubated with shaking (200 rpm, 37°C, 12–16 hours).

Study design.

Primary influent wastewater grab samples were collected from a local wastewater treatment plant in Seattle, WA, that processes 90 million gallons per day (mgd) during the dry season and can process more than 300 mgd during the rainy season,⁴⁰ resulting in matrix variability. Grab samples (7–14 L per carboy) were stored at 4°C until processing (conducted within 72 hours of collection). Varying amounts of Ty21a or Ty2 were seeded to 10 mL 1× phosphate-buffered saline, vortexed (30 seconds), and seeded into varying volumes of wastewater to reach the target concentrations. The final concentration of Ty2 or Ty21a in the seeded wastewater varied depending on the experiment and ranged from approximately 0.001 to 10,000 colony-forming units (CFU) per milliliter (Supplemental Figure 1). Methods tested at the same concentration level used primarily the same initial wastewater matrix with replicates of three or six to enable comparison between the methods. The seeded concentrations were assessed in parallel for each experiment via spread plating of 100 µL of relevant dilutions on LB-Miller agar (Ty21a) or LB-Miller agar with a supplemental aromatic amino acid mix and 50 ng/mL ferrioxamine E (Ty2). The seeded wastewater was thoroughly mixed and then distributed using a peristaltic pump while continuously shaken for processing by 1) filter cartridge, 2) differential centrifugation, 3) grab enrichment, 4) membrane filtration, and 5) Moore swab methods (Figure 1). The methods were evaluated in a laboratory setting, and factors considered included feasibility of performing these methods in other settings and Salmonella Typhi detection by qPCR. Feasibility was based on processing time, equipment, and supplies needed. Active time was defined as the total hands-on time required to conduct an individual method, excluding time in which personnel could complete other tasks. These periods could include incubation, shaking, and filtration.

View Full Size

Workflow of method evaluations. Shaded boxes indicate where samples were collected for DNA extraction. cent. = centrifugation; D = direct; OB = ox bile; qPCR = quantitative polymerase chain reaction; SC = selenite cystine broth; SF = selenite F broth; UPE = universal pre-enrichment broth. Italicized, bolded text indicates method names.

Citation: The American Journal of Tropical Medicine and Hygiene 108, 3; 10.4269/ajtmh.22-0427

• Download Figure
• Download figure as PowerPoint slide

Methods evaluated.

DNA extraction and qPCR.

DNA extraction was performed on the samples using the QIAamp PowerFecal Pro DNA Kit (Qiagen, Hilden, Germany) according to manufacturer’s instructions, with the following modifications. The input was pelleted 1-mL aliquots of the samples (FC1-D, DC-D–50 mL, DC-SF–50 mL, DC-D–1 L, MF1-D, MF1-OB, MF1-SC, MF2-SF, and MS1-SF), pelleted secondary concentrates (FC2-D), or sliced membrane filters (GE-UPE and MS2-UPE). The DNA was eluted in 60 µL and aliquoted into two 30-µL volumes or eluted in 120 µL and aliquoted into three 40-µL volumes and stored at −20°C prior to qPCR.

Samples were analyzed for Ty21a and Ty2 via a qPCR assay targeting the staG gene commonly used for Salmonella Typhi detection in human clinical samples.³⁸ The reaction was carried out with 0.4 µM of the forward (5′-CGCGAAGTCAGAGTCGACATAG-3′) and reverse (5′-AAGACCTCAACGCCGATCAC-3′) primers, 0.15 µM of the probe (5′-FAM-CATTTGTTCTGGAGCAGGCTGACGG-BHQ), and 1× iTaq Universal Probes Supermix (Bio-Rad Laboratories, Inc., Hercules, CA). A DNA extract input volume of 5 µL was used with a 25-µL reaction. Each DNA extract was analyzed undiluted and at a 10-fold dilution to monitor for PCR inhibition effects. Unseeded wastewater controls, no template controls, and a positive control dilution series (standard curve) of Ty21a or Ty2 were analyzed with each qPCR run. Unseeded wastewater controls were positive in 17 of 56 wells throughout the study, with a minimum cycle threshold (Ct) value of 34.1 and maximum Ct value (considered positive) of 39.0. No template controls were negative throughout the course of the study. Standard curve efficiencies averaged 104%, and R² values were > 0.96. The input for the standard curves was prepared by centrifuging (10 minutes, 10,000 ×g) a 1-mL volume of the Ty21a or Ty2 overnight culture harvested during exponential growth (quantified via spread plating of 100 µL of relevant dilutions on LB-Miller agar [Ty21a] or LB-Miller agar with a supplemental aromatic amino acid mix and 50 ng/mL ferrioxamine E [Ty2]), removing the supernatant, and performing DNA extraction as described above. All samples and controls were tested in duplicate or triplicate.

Samples positive for Salmonella Typhi were defined as those that amplified with a Ct of 40 or lower in one or two of the two technical replicates with an appropriately shaped curve. Samples with a Ct > 40 were assumed to be negative because of the potential for spurious artifacts to interfere with detection. The limit of detection (LOD) was determined to be a Ct of 37 (corresponding to approximately 100 CFU/mL of the original culture) for both Ty21a and Ty2 and was defined as 95% samples positive for this assay on this qPCR instrument used (Bio-Rad CFX96 Instrument; Bio-Rad Laboratories, Inc.).⁴² The limit of quantification was determined to be a Ct of 33 (corresponding to approximately 1,000 CFU/mL of the original culture) based on a coefficient of variation below 35%.⁴³ As samples above the LOD but below a Ct of 40 have a higher likelihood of false negatives, samples were determined to be positive if they amplified at a Ct of 40 or lower.

Data analysis.

Methods were primarily evaluated for their rate of Salmonella Typhi positivity. This was determined irrespective of the application of enrichment steps and the sample volume processed, as larger sample volumes do not inherently yield greater positivity. However, the effective volume assayed for the various methods was determined as shown below. Methods were not able to be evaluated for recovery efficiency because of the use of enrichment steps in several methods and the use of Moore swabs (collecting an unknown amount of volume).

Method feasibility.

The feasibility of these methods depends on a variety of factors such as timing, volumes concentrated, supplies and equipment required, and safety (Tables 1 and 2). The active time for each method is < 3 hours. The DC-D–50 mL, DC-SF–50 mL, GE-UPE, and MS1-SF methods each require approximately 60 minutes or less of active personnel time. The total time to yield concentrates for DNA extraction can be obtained for most methods within 48 hours, including active and inactive time. The exception to this timing was for Moore swabs, which require 3–6 days for the concentrate owing to the swab holding time in wastewater and the incubation time. The sample volume concentrated affects the processing time, with larger volumes requiring more processing time (i.e., filter cartridge, DC-D–1 L, and membrane filtration). Procurement of supplies is simple, as all required items are accessible commercially in the United States, though this may vary in other locations. For example, selenite-based enrichment broths can be difficult to obtain in other countries because of chemical safety concerns. These selenite-based enrichment broths are used in the DC-SF–50 mL, MF1-OB, MF1-SC, MF2-SF, and MS1-SF methods. Supply and equipment costs vary between methods and depend heavily on already existing supplies and equipment in other laboratory settings.

Ty21a detection.

Eight methods with 11 variations total were tested for their ability to concentrate wastewater seeded with varying concentrations of Ty21a for detection via qPCR (Table 1).

When seeded at high concentrations (10,000 and 100 CFU/mL), all methods consistently detected Ty21a (100%; Table 3). Four methods maintained a high frequency of positive detection (> 75%) at all concentrations tested: FC1-D, FC2-D, MF1-D, and MS1-SF, whereas other methods yielded fewer positive detections at lower seeded Ty21a concentrations. For DC-D–50 mL, a high frequency of positive detection (100%) was maintained to a seeded concentration of 0.1 CFU/mL compared with DC-SF–50 mL samples, which were minimally detected at or below a seeded concentration of 1 CFU/mL. Positive detection in GE-UPE samples also decreased at a seeded concentration of 1 CFU/mL. MF1-OB samples maintained a high frequency of positive detection to 0.01 CFU/mL compared with MF1-SC samples, which maintained a high frequency of positive detection to 0.1 CFU/mL. MF2-SF and MS2-UPE samples yielded similar results with a high frequency of positive detection down to 0.01 CFU/mL. Of the seven methods that maintained a high rate of positivity (> 75%) at a seeded concentration of 0.01 CFU/mL or lower, four of the methods included an enrichment step.

In general, as the concentration of Ty21a seeded in the samples decreased, the Ct value increased linearly until the lowest Ty21a concentrations were tested, which had similar Ct values to the next lowest concentration (Supplemental Figure 2). Exceptions to this included MF1-OB, which had very low Ct values at a concentration of 0.01 CFU/mL, and MS1-SF and MS-UPE, which had fairly consistent Ct values at all concentrations tested, likely because of the enrichment step used. With a seeded concentration of 10,000 and 100 CFU/mL, all samples yielded Ct values < 40, with less than a 1-log variation in Ct values between replicates for a majority of sample types (82%, 9/11 and 73%, 8/11 for 10,000 and 100 CFU/mL, respectively).

Ty2 detection.

Five methods were further tested with Ty2: FC2-D, DC-D–1 L, MF1-D, MS1-SF, and MS2-UPE. When seeding 0.1 CFU Ty2/mL, four methods yielded high detection rates: FC2-D, DC-D–1 L, MS1-SF, and MS2-UPE (> 89%; Table 4). When the seeding concentration was decreased to 0.01 CFU/mL, DC-D–1 L, MF1-D, MS1-SF, and MS2-UPE had high detection rates (> 67%; Table 4). In general, seeding 0.1 CFU/mL resulted in a lower average Ct value and greater frequency of positivity compared with seeding 0.01 CFU/mL (Supplemental Figure 3).

Ty21a and Ty2 detection.

This study examined the detection and percent positivity of Ty21a by 11 ES concentration methods and Ty2 by five ES concentration methods. Methods were not optimized or evaluated for recovery efficiency, but were rather chosen based on percent positivity and feasibility. As anticipated, when the concentration seeded into the wastewater decreased, the positive detection rate generally decreased. Exceptions to this were seen with decreased recovery of Ty21a at an intermediate concentration in FC1-D, FC2-D, and MS1-SF samples, which could be due to variability between experiments, as the wastewater matrix was collected at different times for each experiment, and variability in the Ty21a strain, as it was attenuated. Typically, methods that assayed larger volumes of the initial sample generally yielded a higher positive detection rate at low-seeded concentrations than did methods that assayed smaller volumes of seeded wastewater. A high percentage of recovery was measured at low Ty21a seeding concentrations for FC1-D and FC2-D methods, which assayed the largest volume of the initial sample (85 and 123 mL, respectively; Table 1), and for MS1-SF samples. The Moore swab methods are inherently non-volumetric. The hold-up volume of the Moore swabs was approximately 60 mL (amount of liquid the Moore swabs absorbed), and they were exposed to 5 L of recirculated sample in this study. Methods that assayed < 1 mL of the original sample included DC-SF–50 mL (minimal detection at 0.01 CFU Ty21a/mL), MF1-OB (minimal detection at 0.001 CFU Ty21a/mL), MF1-SC (no detection at 0.01 CFU Ty21a/mL), and GE-UPE (minimal detection at 1 and 0.1 CFU Ty21a/mL and no detection at 0.01 CFU Ty21a/mL). This information was factored into the decision to exclude these methods from further experiments conducted with Ty2.

For all methods tested with Ty2, when the seeded concentration decreased, the positive detection rate decreased. Additionally, for methods with a calculable volume of the original sample assayed, high volumes processed yielded more positive detection, with DC-D–1 L (9.7 mL) yielding the highest rate of positive detection, followed by FC2-D (123 mL) and MF1-D (2 mL). Finally, where the same assay was used, Ty21a was detected at lower seeded concentrations than Ty2. This could be due to differences in assay sensitivity to Ty21a and Ty2, variability in experiments or wastewater matrix used, differences in the organisms captured by these methods, or differences in the stability of these organisms in the wastewater matrix throughout processing.

Method feasibility and use case scenarios.

Method selection depends on field logistics, laboratory constraints, project design, and budgetary considerations; an appropriate ES method will operate within these confines while maintaining effective performance. Local field conditions and infrastructure impact appropriate surveillance sites (e.g., sewer, wastewater treatment plant pumping station, river, pit latrine) and therefore the appropriate matrix. These matrices will vary by available sample volume, total solids, and solid characteristics (e.g., sediments, debris), which in turn will inform appropriate sampling and concentration method selection. For example, matrices with high solids content may be a challenge for filter cartridge and membrane filtration methods because of filter clogging, whereas methods such as GE-UPE, differential centrifugation, and Moore swabs are able to process samples with high solids content. Filter cartridge methods rely on the adsorption of negatively charged bacteria onto positively charged filters with a pore size of 2–3 µm; thus, these methods may be less applicable in turbid waters (e.g., high-strength wastewater or pit latrine waste) as less volume may be able to be filtered, which will affect recovery. Additionally, sampling sites with very low flows may not be applicable for methods that process large volumes, such as filter cartridge, DC-D–1 L, or membrane filtration.

Site access and field-worker safety are critical considerations for sample collection and in-field processing. If sample shipment between the collection site and processing laboratory is required, then this may impact the sample volume (requiring smaller volume samples, as with DC-D–50 mL, DC-SF–5 mL, or GE-UPE), the need to perform primary concentration at the field site (e.g., with the filter cartridge methods or the in-field version of Moore swabs), and/or the ability to maintain sample integrity over the period of transport. It is important to note that the Moore swabs methods tested in this evaluation were not representative of what happens in their intended use case, as the experimental Moore swab was held in a recirculating system with multiple exposures to the same seeded wastewater rather than placed in a drainage with exposure only to new wastewater throughout the holding period.

Available laboratory equipment and supplies, physical space, and personnel time also impact the ability to conduct methods. For example, although the filter cartridge methods use a commercialized kit, making procurement simple, they also require a large centrifuge and shaking table. Centrifuge capacity is also a challenge for DC-D–1 L, particularly if the centrifuge does not contain a cooling mechanism or if multiple samples must be processed in 1 day. Access to a house vacuum or strong vacuum pump could also be a challenge preventing use of GE-UPE, membrane filtration methods, and MS2-UPE. Laboratorian safety is critical when choosing an appropriate ES method. Both selenite cystine broth (SC) (used for MF1-OB and MF1-SC) and SF (used for DC-SF–50 mL, MF2-SF, and MS1-SF) broths must be prepared and used under a biosafety cabinet capable of chemical protection (e.g., Class II B2) because of their acute toxicity and teratogenicity. Selenite-based enrichment broths cannot be autoclaved and must be disposed of as hazardous chemical waste because of their aquatic toxicity. Therefore, these chemical hazards are important considerations in settings where selenite-based compounds cannot be contained or disposed of appropriately.

Finally, study design, time to results, and associated budgetary considerations necessarily impact method selection. In cases where results are needed rapidly, filter cartridge methods, DC-D–50 mL, DC-D–1 L, or MF1-D may be ideal, as results can be obtained in < 24 hours as no overnight incubation steps are required. This combination of field, laboratory, study, and performance considerations results in a complex decision tree where one outcome is not appropriate for all applications. Thus, it is necessary to have the flexibility to select an appropriate method that meets all or a majority of these needs. Here, different use case scenarios are outlined with potentially appropriate methods (Table 5).

Limitations.

Limitations of this study include variability between and within experiments, the number of replicates for different methods and seeding levels evaluated, the assay chosen for detection, and the methods evaluated. Variability between experiments could result from the wastewater matrix used. Although the background matrix was collected from the same local wastewater treatment plant for every experiment, the experiments were conducted over a 1-year time frame owing to laboratory capacity to process samples. Additionally, wastewater could not be collected once and stored for all experiments, as wastewater characteristics change with storage time and limited storage space was available. Thus, there may have been variability in the wastewater matrix composition due to precipitation events prior to or during certain sample collections, as the treatment plant is a combined sewer overflow system. Additionally, although glycerol stocks of the same initial culture of Ty21a or Ty2 were used to prepare the organisms for seeding and growth curves were determined, there was minor variation in the length of the overnight cultures, potentially introducing variability in the anticipated amount of Ty21a or Ty2 seeded. Overnight cultures were quantified via spread plating, with differences from the anticipated amount ranging from 8% to 156% for Ty21a and 11% to 498% for Ty2. There may also have been variability within an experiment, as it was difficult to fully homogenize large sample volumes. The effect of this may have been amplified at low seeding levels. Although the extraction method used was consistent across all sample types, the extraction efficiency of the different sample types may have varied, though this was not evaluated in this study. Finally, when SF and SC broths were used, the length of the overnight culture in the selenite broths varied from 12 to 18 hours and 20 to 21 hours, respectively. Longer enrichment in selenite-based media can overtax the media, leading to breakthrough of organisms other than the target, Salmonella Typhi.

Variability in the number of replicates per method and per experiment makes comparisons across methods and seeding levels challenging, and only a limited number of replicates were generated for the methods evaluated. Differences in the number of replicates for the different methods were due to the complexity and feasibility of processing multiple replicates using a certain method, as some methods required less wastewater or personnel time than others. For example, the large variability in Ct values seen at 0.01 CFU Ty2/mL compared with 0.1 CFU Ty2/mL with DC-D–1 L could be due to differences in the number of experiments (three were conducted at 0.01 CFU/mL and one conducted at 0.1 CFU/mL), the number of replicates, and challenges with homogenizing a low concentration of Ty2 throughout a sample. Additionally, we expected to see large variation between replicates at low concentrations that approached the assay limit of detection.

This study used the Nga et al.³⁸ primer and probe set for detection of Salmonella Typhi. This assay was originally designed for clinical samples and was recently applied for detection in environmental samples. However, wastewater contains a variety of organisms as well as both known and unknown DNA in the samples, which could interfere with this assay and result in cross reactivity.⁴⁴ The ability of the concentration methods to detect Salmonella Typhi using this assay may also vary, as some methods concentrate larger volumes, resulting in greater concentration of potential qPCR inhibitors. To minimize these effects on reported results, undiluted and 10-fold sample dilutions were assayed to screen for evidence of inhibitors. Future studies should examine improved qPCR assays for ES samples specifically as well as digital qPCR assays. Archived samples from this study could be retested with the optimized method.

Finally, this study evaluated multiple Salmonella Typhi ES methods, though some other ES methods were not included. For example, additional ES methods such as dead-end ultrafiltration and hollow fiber ultrafiltration were not included in this study because of time and resource constraints. Future studies should expand on the types of methods evaluated.

This study evaluated eight methods with 12 formats for their applicability to conduct ES using Salmonella Typhi Ty21a and Ty2. Results suggest that all methods tested can be successful at concentrating Salmonella Typhi for subsequent detection by qPCR, although each method has its own strengths and weaknesses, including the Salmonella Typhi concentrations for which they are applicable. These factors should be considered when identifying a method for Salmonella Typhi ES and will greatly depend on the use case planned. Future studies could benefit from examining additional ES methods not used here and conducting side-by-side evaluations with field samples.